Quantitative Analysis Using Statistics for Forecasting and Validity Testing. Course #6300/QAS6300 Course Material

Size: px
Start display at page:

Download "Quantitative Analysis Using Statistics for Forecasting and Validity Testing. Course #6300/QAS6300 Course Material"

Transcription

1 Quantitative Analysis Using Statistics for Forecasting and Validity Testing Course #6300/QAS6300 Course Material

2 Table of Contents Page Chapter 1: Decision Making With Statistics and Forecasting I. How to Use Basic Statistics 1-1 II. Using Forecasting Techniques 1-5 III. Regression Analysis for Sales and Earnings Projections 1-14 IV. What Statistics to Look for in Regression Analysis 1-19 V. Measuring the Accuracy of Forecasts 1-6 VI. How to Use a Computer Statistical Package for Multiple Regressions 1-7 Review Questions & Solutions 1-36 Chapter : Making Use of Quantitative Decision Making I. Decision Making Under Risk, Certainty, and Conflict -1 II. Decision Making Under Uncertain Conditions -4 III. Decision Theory -6 IV. Linear Programming and Shadow Prices -9 V. Learning Curve -15 VI. Inventory Planning and Control -17 VII. Queuing (Waiting Line) Models -6 Review Questions & Solutions -30 Glossary Index Table of Contents

3 Chapter 1: Decision Making With Statistics and Forecasting Upon completion of this chapter, you will be able to: List the types of a mean and illustrate how they are used. Differentiate between moving averages and exponential smoothing. Discuss regression analysis and its applications in business. Use regression for trend analysis. List types of statistics to look for in regression analysis. Measure forecasting errors. Choose the best forecasting equation. Decision makers find themselves in many situations in which large volumes of data need to be analyzed. These data could be sales figures, income, or a multitude of other information. The data could be used for a variety of purposes, including risk analysis, figuring return on investments, or other financial decisions. Effective use of statistics and forecasting techniques will prove necessary as a company grows. I. How to Use Basic Statistics The most commonly used statistics that describe characteristics of data are the mean and the standard deviation. A. TYPES OF MEAN The mean gives an average (or central) value of a data set. Three such means are common: Arithmetic mean Weighted mean Geometric mean B. ARITHMETIC MEAN The arithmetic mean is a simple average. To find it, sum the values of a data set and divide by the number of data entries: where x = the arithmetic mean (called x-bar) x = the data values n = number of observations x x = n Decision Making With 1-1 Statistics and Forecasting

4 Example 1 John Jay Lamp Company has a revolving credit agreement with a local bank. Last year, the loan showed the following month-end balances: January $18,500 February 1,000 March 17,600 April 3,00 May 18,600 June 4,500 July 60,000 August 40,000 September 5,850 October 33,100 November 41,000 December 8,400 The mean monthly balance is computed as follows: $18,500 + $1,000 + $17,600 + $3,00 + $18,600 + $4,500 + $60,000 + $40,000 + $5,850 + $33,100 + $41,000 + $8,400 1 = $351,750 1 = $9,31.50 C. WEIGHTED MEAN When the data values have different degrees of importance or frequency, a weighted mean helps account for this. The formula for a weighted mean is: Weighted mean = ( w)(x) where w = the weight assigned to each observation, expressed as a percentage or relative frequency. Example Company J uses three grades of labor to produce a finished product as follows: Labor Hours per Grade of Labor Unit of Labor Hourly Wages (x) Skilled 6 $10.00 Semiskilled Unskilled Decision Making With 1- Statistics and Forecasting

5 The arithmetic mean (average cost) of labor per hour for this product can be computed as follows: Arithmetic mean = $ $ $ = $8.00 per hour However, this formula assumes that each grade of labor was used in equal amounts, and this is not the case. To calculate the average cost of labor per hour correctly, the weighted average should be computed as follows: Weighted mean = $10.00(6/10) + $8.00(3/10) + $6.00(1/10) = $9.00 per hour Note: The weights equal the proportion of the total labor required to produce the product. D. GEOMETRIC (COMPOUND) MEAN Sometimes quantities change over a period of time; for example, the rate of return on investment or rate of growth in earnings over a period of years. In such cases, employ the geometric mean, which uses the average rate or percentage of change. where Geometric (compound) mean = n (1 + x )(1 + x )...(1+ x ) 1 x = the rate of change (in percent) n = number of periods Example 3 1 n A stock doubles during one period and then depreciates back to the original price, as shown in the following table. Dividend income (current income) is non-existent. Time Periods t = 0 t = 1 t = Price (end of period) $80 $160 $80 HPR 100% -50% The rate of return for periods 1 and are computed as follows: Period 1 (t=1) = $ 160 $80 $80 = $80 $80 = 100% Period (t=) = $ 80 $160 $160 = $80 $160 = -50% The rate of return is the average of 100% and -50%, or 5%: 100% + ( 50%) = 5% Decision Making With 1-3 Statistics and Forecasting

6 However, the stock bought for $40 and sold for the same price two periods later did not earn 5%; it earned zero. This can be illustrated by determining the compound average return. n =, x 1 = 100% = 1, and x = -50% = Then, Geometric (compound) mean return = (1 + x )(1 + x ) 1 = (1+1)(1 0.5) 1 where x 1 = 100% or 1 x = -50% or -0.5 E. STANDARD DEVIATION = ()(0.5) 1 = 1 1 = 1 1 = 0 1 The standard deviation measures the extent to which data spread out or disperse. Managers can make important inferences from past data with this statistic, such as when measuring the risk of purchasing a financial asset. The standard deviation, denoted with the Greek letter σ (read as sigma), is defined as follows: σ = where x is the mean (arithmetic average). (x x) n 1 Calculate the standard deviation using these five steps: 1. Subtract the mean from each element of the data.. Square each of the differences obtained in step Add together all the squared differences. 4. Divide the sum of all the squared differences by the number of values minus one. 5. Take the square root of the quotient obtained in step 4. Example 4 One and a half years of quarterly returns are listed below for ABC Mutual Fund. Time period x (x x) (x x) 1 10% Decision Making With 1-4 Statistics and Forecasting

7 The mean return and standard deviation over this period are computed as follows: x = 60/6 = 10% σ = (x x) = n = 130 = 11.40% ABC Fund has returned on the average 10 percent over the last six quarters and the variability about its average return was 11.40%. The high standard deviation (11.40%) relative to the average return (10%) indicates that the fund is very risky. II. Using Forecasting Techniques A. USING MOVING AVERAGES With the moving average, use the most recent observations (n) to calculate an average. Then, use this as the forecast for the next period. Moving averages are updated as new data are received. Note: Choose the number of periods to use on the basis of the relative importance attached to old versus current data. Example 5 Assume that the marketing manager has the following sales data. Date Actual Sales (Y t ) Jan In order for the marketing manager to predict the sales for the seventh and eighth days of January, she must pick the number of observations to be averaged. She used two possibilities: a six-day and a three-day period. Case 1 where Y' = predicted Y 7 = = Y 8 = = Decision Making With 1-5 Statistics and Forecasting

8 Case Y 7 = = Y 8 = = Predicted Sales (Y' t ) Date Actual Sales Case 1 Case Jan In terms of the relative importance of new versus old data, in Case 1, the old data received a weight of 5/6 and current data 1/6. In Case, the old data received a weight of only /3, while current data received 1/3 weight. Thus, the marketing manager's choice of the number of periods to use in a moving average is a measure of the relative importance attached to old versus current data. B. ADVANTAGES AND DISADVANTAGES OF MOVING AVERAGES The moving average is simple to use and easy to understand. However, it has three flaws. It requires the retention of a great deal of data and moving the data from forecast period to forecast period. All data in the sample are weighted equally. If more recent data are more valid than older data, it cannot be differently weighted with a moving average. It cannot explain the underlying causal relationships among variables. The forecasting method known as exponential smoothing circumvents these disadvantages. Decision Making With 1-6 Statistics and Forecasting

9 C. THE BASIS FOR EXPONENTIAL SMOOTHING Exponential smoothing is a popular technique for short-run forecasting. It uses a weighted average of past data as the basis of the forecast. The procedure assumes the future is more dependent upon the recent past than on the distant past and thus gives heaviest weight to more recent data and smaller weight to those of the more distant past. The method is most effective when there is randomness and no seasonal fluctuations. However, the method does not include industrial or economic factors, such as market conditions, prices, or competitors' actions. D. THE EXPONENTIAL SMOOTHING MODEL The formula for exponential smoothing is: or Y' t+1 = αy t + (1 α)y' t Y' new = αy old + (1 α)y' old where Y' new = exponentially smoothed average to be used as the forecast Y old = most recent actual data Y' old = most recent smoothed forecast α = smoothing constant The higher the α, the higher the weight given to the more recent data. Example 6 YSY provides the following sales data: Time period (t) Actual sales (1000)(Y t ) 1 $ To initialize the exponential smoothing process, there must be an initial forecast. The first smoothed forecast to be used can be: Decision Making With 1-7 Statistics and Forecasting

10 1. First actual observations, or. An average of the actual data for a few periods. The manager decides to use a six-period average as the initial forecast Y' 7 with a smoothing constant of = Then Y' 7 = (Y 1 + Y + Y 3 + Y 4 + Y 5 + Y 6 )/6 = ( )/6 = 63 Note that Y 7 = 70. Then, Y' 8 is computed as follows: Y' 8 = αy 7 + (1 α)y' 7 = (0.40)(70) + (0.60)(63) = = Y' 9 and Y' 10 may be similarly calculated: Y' 9 = αy 8 + (1 α)y' 8 = (0.40)(74) + (0.60)(65.80) = = Y' 10 = αy 9 + (1 α)y' 9 = (0.40)(6) + (0.60)(69.08) = = 66.5 By using the same procedure, the values of Y' 11, Y' 1, Y' 13, Y' 14, and Y' 15 can be calculated. The following shows a comparison between the actual sales and predicted sales by the exponential smoothing method. COMPARISON OF ACTUAL SALES AND PREDICTED SALES Time period (t) Actual sales (Y t ) Predicted sales (Y' t ) Difference (Y t - Y' t ) Difference (Y t - Y' t ) 1 $ Decision Making With 1-8 Statistics and Forecasting

11 E. DETERMINING THE BEST SMOOTHING CONSTANT A higher or lower smoothing constant (α) can be used in order to adjust a prediction to large fluctuations in the data series. For example, if the forecast is slow in reacting to increased sales (if the difference is negative), try a higher value. For practical purposes, the optimal α may be picked by minimizing what is known as the mean squared error (MSE), which is the average sum of the variations between the historical data and forecast values for the corresponding periods. MSE is computed as follows: MSE = Σ (Y t Y' t ) / (n i) where i = the number of observations used to determine the initial forecast (in our example, i = 6). In the previous example, i = 6 and Tip: Try to select the α that minimizes MSE. F. HOW A COMPUTER CAN HELP MSE = / (15 6) = / 9 = Managers are often confronted with complex problems requiring large samples of data. This will require trying different values of α for exponential smoothing. A computer can assist in these instances. Virtually all forecasting software have an exponential smoothing routine. To demonstrate, consider the following data. Time Actual Sales Period (in Thousands of Dollars) Figure 1-1 is a printout of an exponential smoothing program. The best α for this particular example is 0.9, since it gives the least MSE. Decision Making With 1-9 Statistics and Forecasting

12 FIGURE 1-1 EXPONENTIAL SMOOTHING PROGRAM Decision Making With 1-10 Statistics and Forecasting

13 Decision Making With 1-11 Statistics and Forecasting

14 Decision Making With 1-1 Statistics and Forecasting

15 Decision Making With 1-13 Statistics and Forecasting

16 III. Regression Analysis for Sales and Earnings Projections A. DEFINING REGRESSION ANALYSIS Regression analysis is a statistical procedure for mathematically estimating the average relationship between the dependent variable and the independent variable(s). Simple regression involves one independent variable, such as price or advertising in a demand function, whereas multiple regression involves two or more variables, such as price and advertising together. Simple (linear) regression is defined by the following equation: Y = a + bx where Y = dependent variable X = independent (explanatory) variable a = a constant or Y intercept of regression line b = the slope of the regression line B. USING THE METHOD OF LEAST SQUARES The method of least squares attempts to find a line of best fit for the graph of a regression equation. To better explain this, define error, or u, as the difference between the observed and estimated values of sales or earnings: u = Y - Y' where Y = observed value of the dependent variable Y' = estimated value based on Y' = a + bx The least-squares criterion requires that the line of best fit be such that the sum of the squares of the errors (the vertical distance in Figure 1- from the observed data points to the line) is a minimum, i.e., Minimum: Σu = Σ(Y-a-bX) Decision Making With 1-14 Statistics and Forecasting

17 FIGURE 1- ACTUAL (Y) VERSUS ESTIMATED (Y') Note: The objective of a scatter diagram is to demonstrate correlations. Each observation is represented by a dot on a graph corresponding to a particular value of X (the independent variable) and Y (the dependent variable). Also, make sure that a linear (straight-line) relationship existed between Y and X in the past sample. Using differential calculus yields the following equations, called normal equations: Solving the equations for b and a yields b = n ΣY = na + bσx ΣXY = aσx + bσx n a = XY X Y bx ( X)( Y) ( X) where Y = Y and X = n n X Decision Making With 1-15 Statistics and Forecasting

18 Example 7 To illustrate the computations of b and a, we will refer to the data in Table 1-1. All the sums required are computed and shown below. TABLE 1-1 ORIGINAL DATA AND COMPUTED SUMS Advertising X (000) Sales Y (000) XY X Y $ $174 $5 3,414,79 4,359 From the table above: ΣX = 174 ΣY = 5 ΣXY = 3,414 ΣX =,79 X = ΣX/n = 174/1 = 14.5 Y= ΣY/n = 5/1 = Substituting these values into the formula for b first: a = n b = n XY X ( X)( Y) ( X) (1)(3,414) (174)(5) = = (1)(,79) (174) Y bx = (0.563)(14.5) = = Thus, Y' = X 1,818 3,8 = Assume that an advertising budget of $10 will be expended for next year. The projected sales for the next year would be computed as follows: Y' = X = (10) = $ Note: ΣY is not used here but rather is computed for r-squared (R ). Decision Making With 1-16 Statistics and Forecasting

19 C. USING TREND ANALYSIS Trends are the general upward or downward movements of the average over time. These movements may require many years of data to determine or describe. They can be described by a straight line or a curve. The basic forces underlying the trend include technological advances, productivity changes, inflation, and population change. Trend analysis is a special type of simple regression. This method involves a regression whereby a trend line is fitted to a time series of data. The linear trend line equation can be shown as where t = time. Y = a + bt The formula for the coefficients a and b are essentially the same as the cases for simple regression. However, for regression purposes, a time period can be given a number so that Σt = 0. When there is an odd number of periods, the period in the middle is assigned a zero value. If there is an even number, then -1 and +1 are assigned the two periods in the middle so that again Σt = 0. With Σt = 0, the formula for b and a reduces to the following: Example 8 Case 1 (odd number) Case (even number) In each case Σt = 0. b = a = n n ty t Y n 0X1 0X 0X3 0X4 0X5 t = X1 0X 0X3 0X4 0X5 0X6 t = Decision Making With 1-17 Statistics and Forecasting

20 Example 9 Consider ABC Company, whose historical sales follows. Year Sales (in millions) 0X1 $ 10 0X 1 0X3 13 0X4 16 0X5 17 Since the company has five years' data, which is an odd number, the year in the middle is assigned a zero value. Year t Sales(in ty t Y millions)(y) 0X1 - $ X X X X b = (5)(18) 5(10) = 90/50 = a= = Therefore, the estimated trend equation is: Y' = $ $1.8t To project 0X6 sales, assign +3 to the t value for the year 0X6. Y' = $ $1.8(3) = $19 A summary of the four forecasting methods described in this chapter is provided in Figure 1-3. Use it as a guide for determining which method is best for a specific circumstance. Decision Making With 1-18 Statistics and Forecasting

21 FIGURE 1-3 SUMMARY OF MORE COMMONLY USED FORECASTING METHODS IV. What Statistics to Look for in Regression Analysis A variety of statistics can be used to determine the accuracy and reliability of the regression results. Three such statistics are: 1. Correlation coefficient (R) and coefficient of determination(r ). Standard error of the estimate (S e ) and prediction confidence interval 3. Standard error of the regression coefficient (S b ) and t-statistic Decision Making With 1-19 Statistics and Forecasting

22 A. MEASURING THE GOODNESS OF FIT FOR THE REGRESSION EQUATION The correlation coefficient R measures the degree of correlation between Y and X. The range of values it takes on is between -1 and +1. More widely used, however, is the coefficient of determination, designated R (read as r-squared). R identifies how good the estimated regression equation is. In other words, it is a measure of "goodness of fit" in the regression. The higher the R, the more confidence there is in the estimated equation. The coefficient of determination represents the proportion of the total variation in Y that is explained by the regression equation. It has the range of values between 0 and 1. Example 10 The statement "Sales is a function of advertising expenditure with R = 70 percent," can be interpreted as "70 percent of the total variation of sales is explained by the regression equation or the change in advertising, and the remaining 30 percent is accounted for by something other than advertising, such as price and income." The coefficient of determination is computed as: R = 1 (Y Y ) (Y Y) In a simple regression situation, this short-cut formula can be used: R = n X [ n XY ( X)( Y) ] ( ) ( ) X n Y Y Comparing this formula with the one for b, it is evident that the only additional information needed to compute R is ΣY. Example 11 Refer to Table 1-1 shown in Example 7. With Y, one can compute R using the short-cut formula as follows: R (1,818) 3,305,14 = = = [3,8][(1)(4,359) (5) ] (3,8)(5,308 50,65) 3,305,14 (3,8)(1,683) = 3,305,14 5,43,74 = = 60.84% About percent of the total variation in sales is explained by advertising and the remaining percent is still unexplained. A relatively low R indicates that there is a lot of room for improvement in the estimated forecasting formula (Y' = $ $0.563X). Price or a combination of advertising and price might improve R. Decision Making With 1-0 Statistics and Forecasting

23 A low R is an indication that the model is inadequate for explaining the y variable. The general causes for this problem are: 1. Use of a wrong functional form.. Poor choice of an X variable as the predictor. 3. The omission of some important variable or variables from the model. B. MEASURING THE ACCURACY OF MANAGEMENT PREDICTIONS The accuracy of management predictions can be determined by using the standard error of the estimate, designated S e, and defined as the standard deviation of the regression. The standard error of the estimate represents the variance of actual observations from the regression line. S e (Y Y ) Y a Y b XY = = n n This statistic can be used to gain some idea of the accuracy of a prediction. Example 1 Using the example data from Table 1-1, S e is calculated as: S e Y a Y b XY = n = 4,359 ( )(5) (0.563)(3,414) 1 =.3436 = Suppose an organization desires to make a prediction regarding an individual Y value, such as a prediction about the sales when an advertising expense = $10. Usually, it helps to have some objective measure of the confidence that can be placed in a prediction. One such measure is a confidence (or prediction) interval constructed for Y. A confidence interval for a predicted Y, given a value for X, can be constructed in the following manner: Y ± ts e 1+ 1 n + (X X p X) ( X) n where Y' = the predicted value of Y given a value for X X p = the value of independent variable used as the basis for prediction Note: t is the critical value for the level of significance employed. For example, for a significant level of 0.05 (which is equivalent to a 95% confidence level in a two-tailed test), the critical value of t for 10 degrees of freedom is.8 (See Table 1-3). As can be seen, the confidence interval is the linear distance bounded by limits on either side of the prediction. Decision Making With 1-1 Statistics and Forecasting

24 When a 95 percent confidence interval for the prediction is desired, the range for the prediction, given an advertising expense of $10, would be between $10, and $1,836.10, as the following calculation shows. Note that from Example 7, Y' = $ The confidence interval is therefore established as follows: $ ± (.8)(.3436) = $ ± (.8)(.3436) (1.0764) = $ ± ( ) + (174),79 1 which means the range for the prediction, given an advertising expense of $10, would be between $ and $ Note that $ = $ and $ = $ C. TESTING THE APPROPRIATENESS OF THE REGRESSION COEFFICIENT The standard error of the regression coefficient, designated S b, and the t-statistic are closely related. S b is calculated as: S b = S e (X X) or in short-cut form, S b = X S e X X S b gives an estimate of the range where the true coefficient will "actually" fall. The t-statistic (or t-value) is a measure of the statistical significance of an independent variable X in explaining the dependent variable Y. It is determined by dividing the estimated regression coefficient b by its standard error S b. It is then compared with the table t-value (See Table 1-3). Thus, the t-statistic measures how many standard errors the coefficient is away from zero. Rule of thumb: Any t-value greater than + or less than - is acceptable. The higher the t-value, the greater the confidence we have in the coefficient as a predictor. Low t-values are indications of low reliability of the predictive power of that coefficient. Example 13 The S b for our example is: S X S X X.3436,79 (14.5)(174).3436,79,53 e b = = = = Decision Making With 1- Statistics and Forecasting

25 Thus, t-statistic = b = S b = 3.94 Since t = 3.94 >, it can be concluded that the b coefficient is statistically significant. As was indicated previously, the table's critical value (cut-off value) for 10 degrees of freedom is.8 (from Table 1-3 at the end of the chapter). Review: (1) t-statistic is more relevant to multiple regressions which have more than one b. () R indicates how good the forest (overall fit) is while t-statistic tells you how good an individual tree (an independent variable) is. In summary: The table t value (retrieved from Table 1-3), based on a degree of freedom and a level of significance, is used: (1) To set the prediction range upper and lower limits for the predicted value of the dependent variable. () To set the confidence range for regression coefficients. (3) As a cutoff value for the t-test. Figure 1-4 shows an Excel regression output that contains the statistics discussed so far. FIGURE 1-4 EXCEL REGRESSION OUTPUT SUMMARY OUTPUT Regression Statistics Multiple R R Square (R ) Adjusted R Square Standard Error.3436 (S e ) Observations 1 ANOVA df SS MS Significance F Regression Residual Total Coefficients Standard Error (S b ) t Stat Lower 95% Upper 95% Upper 95.0% Intercept Advertising (1) R-squared (R ) = = 60.84% () Standard error of the estimate (S e ) =.3436 (3) Standard error of the coefficient (S b ) = (4) t-value = 3.94 All of the above data are the same as the data obtained manually. Decision Making With 1-3 Statistics and Forecasting

26 E. STATISTICS TO LOOK FOR IN MULTIPLE REGRESSIONS Multiple regressions are used to find the overall association between the dependent variable and explanatory independent variables. You need to take note of the following statistics when doing multiple regressions: t-statistic R-bar squared ( R ) and F-statistic Multicollinearity Autocorrelation (or serial correlation) F. WHAT THE T-STATISTIC INDICATES Even though the t-statistic was discussed in the previous section, it is more valid in multiple regression. The t-statistic shows the significance of each explanatory variable in predicting the dependent variable. In a multiple regression situation, the t-statistic is defined as t-statistic = b S i b i where i = ith independent variable. Rule of thumb: It is desirable to have as large (either positive or negative) a t-statistic as possible for each independent variable. Generally, a t-statistic greater than +.0 or less than -.0 is acceptable. Explanatory variables with low t-value can usually be eliminated from the regression without substantially decreasing R or increasing the standard error of the regression. Table 1- provides t-values for a specified level of significance and degrees of freedom. G. MEASURING GOODNESS OF FIT For multiple regressions, goodness of fit is best represented by R-bar squared ( R ): R = 1 (1 R n 1 ) n k where n = the number of observations k = the number of coefficients to be estimated An alternative test of the overall significance of a regression equation is the F-test. The F-statistic is defined as: (Y Y) / k Explained variation/k F = = (Y Y ) /(n k 1) Unexplained variation/(n k 1) If the F-statistic is greater than the table value, it is concluded that the regression equation is statistically significant in overall terms. Note: Virtually all computer programs for regression analysis show R and the F-statistic. Decision Making With 1-4 Statistics and Forecasting

27 H. HOW TO BE SURE THE INDEPENDENT VARIABLES ARE UNRELATED When using more than one independent variable in a regression equation, there is sometimes a high correlation between the independent variables themselves. Multicollinearity occurs when these variables interfere with each other. It is a pitfall because the equations with multicollinearity may produce spurious forecasts. Multicollinearity can be recognized when: The t-statistics of two seemingly important independent variables are low. The estimated coefficients on explanatory variables have the opposite sign from that which would logically be expected. There are two ways to circumvent the problem of multicollinearity: One of the highly correlated variables may be dropped from the regression. The structure of the equation may be changed using one of the following methods: Divide both the left and right-hand side variables by some series that will leave the basic economic logic but remove multicollinearity. Estimate the equation on a first-difference basis. Combine the collinear variables into a new variable, which is their weighted sum. I. AUTOCORRELATION (SERIAL CORRELATION) Autocorrelation is a problem often encountered in regression analysis. It occurs where there is a correlation between successive errors, which indicates that observations are not independent. The Durbin-Watson statistic provides the standard test for autocorrelation. Table 1- at the end of the chapter provides the values of the Durbin- Watson statistic for specified sample sizes and explanatory variables. Table 1- gives the significance points for d L and d U for tests on the autocorrelation of residuals (when no explanatory variable is a lagged endogenous variable). The number of explanatory variables, K excludes the constant term. Generally speaking, Durbin-Watson Statistic Between 1.5 and.5 Below 1.5 Above.5 Autocorrelation No autocorrelation Positive Autocorrelation Negative Autocorrelation Autocorrelation usually indicates that an important part of the variation of the dependent variable has not been explained. The best solution to this problem is to search for other explanatory variables to include in the regression equation. An example showing applications of all the tests discussed in this section can be found in section VI. Decision Making With 1-5 Statistics and Forecasting

28 V. Measuring the Accuracy of Forecasts The performance of a forecast should be checked against its own record or against that of other forecasts. There are various statistical measures that can be used to measure performance of the model. Of course, the performance is measured in terms of forecasting error, where error is defined as the difference between a predicted value and the actual result. Error (e) = Actual (A) Forecast (F) A. MEASURES THAT ARE COMMONLY USED FOR SUMMARIZING ERRORS Two measures are commonly used for summarizing historical errors: the mean absolute deviation (MAD) and the mean squared error (MSE). The formulas used to calculate MAD and MSE are: MAD = Σ e / n MSE = Σ e / (n - 1) The following example illustrates the computation of MAD and MSE. Example 14 Sales data of a microwave oven manufacturer are given below. Using the figures, Period Actual (A) Forecast (F) e (A-F) e e MAD = Σ e /n = /8 =.75 MSE = Σ e / (n - 1)= 76/7 = One way these measures are used is to evaluate forecasting ability of alternative forecasting methods. For example, using either MAD or MSE, a forecaster could compare the results of exponential smoothing with alphas and elect the one that performed best in terms of the lowest MAD or MSE for a given set of data. Also, it can help select the best initial forecast value for exponential smoothing. Note: Forecasting control can be accomplished by comparing forecasting errors to predetermined values or limits. Errors that fall within the limits would be acceptable while errors outside of the limits would signal that corrective action is required. Decision Making With 1-6 Statistics and Forecasting

29 B. CHOOSING THE BEST FORECASTING EQUATION Choosing among alternative forecasting equations basically involves two steps. The first step is to eliminate the obviously flawed equations. The second is to select the best among the remaining contenders. How to Eliminate Flawed Equations 1. Does the equation make sense? Equations that do not make sense intuitively or from a theoretical standpoint must be eliminated.. Does the equation have explanatory variables with low t-statistics? These equations should be reestimated or dropped in favor of equations in which all independent variables are significant. This test will eliminate equations where multicollinearity is a problem. 3. How about a low R? The R can be used to rank the remaining equations in order to select the best candidates. A low R could mean: A wrong functional form was fitted An important explanatory variable is missing Other combinations of explanatory variables might be more desirable How to Choose the Best Equation 1. The Durbin-Watson statistic. Given equations that survive all previous tests, the equation with the Durbin-Watson statistic closest to.0 can be a basis for selection.. Forecasting accuracy. Examining the forecasting performance of the equations is essential for selecting one equation from those that have not been eliminated. The equation whose prediction accuracy is best in terms of measures of forecasting errors, such as MAD, MSE, RMSE, or MPE, generally provides the best basis for forecasting. It is important to note that neither Lotus 1--3 nor Quattro Pro calculate many statistics such as R-bar squared ( R ), F-statistic, and Durbin-Watson statistic. It is necessary to use regression packages, such as Statistical Analysis System (SAS), MINITAB, and Statistical Packages for Social Scientists (SPSS), to name a few. These packages all have PC versions. VI. How to Use a Computer Statistical Package for Multiple Regressions A. COMPUTER HELP FOR REGRESSION ANALYSES Software packages can greatly assist decision makers and forecasters with a variety of statistical analyses. B. COMPUTER HANDLING OF MULTIPLE REGRESSION An example of how a computer handles multiple regression follows. Remember that each software package is a little different. This example uses SPSS, one of the most popular programs. In Figure 1-5 is a computer listing containing the input data and output results using three independent variables. Illustrative comments have also been added where applicable. Decision Making With 1-7 Statistics and Forecasting

30 Example 15 Stanton Consumer Products Corporation wishes to develop a forecasting model for its dryer sales by using multiple regression analysis. The marketing department has prepared the following sample data using three independent variables: sales of washers, disposable income, and savings. Sales of Disposable Sales of Washers Income Savings Dryers Month (X 1 ) (X ) (X 3 ) (Y) January $45,000 $16,000 $71,000 $9,000 February 4,000 14,000 70,000 4,000 March 44,000 15,000 7,000 7,000 April 45,000 13,000 71,000 5,000 May 43,000 13,000 75,000 6,000 June 46,000 14,000 74,000 8,000 July 44,000 16,000 76,000 30,000 August 45,000 16,000 69,000 8,000 September 44,000 15,000 74,000 8,000 October 43,000 15,000 73,000 7,000 FIGURE 1-5 SPSS REGRESSION OUTPUT Variables Entered/Removed b Model 1 Variables Entered SAVINGS, Variables Removed Method sales, INCOME a. Enter a. All requested variables entered. b. Dependent Variable: SALESDRY Model Summary b Std. Error Model R R Square Adjusted R Square of the Estimate Durbin-Watson 1.99 a a. Predictors: (Constant), SAVINGS, sales, INCOME b. Dependent Variable: SALESDRY Decision Making With 1-8 Statistics and Forecasting

31 Unstandardized Coefficients Coefficients a Standardi zed Coefficien ts Model B Std. Error Beta t Sig. 1 (Constant) sales INCOME SAVINGS a. Dependent Variable: SALESDRY 1. The forecasting equation. The SPSS output is Y' = -45, X X X 3 Suppose that in November the company expects X 1 = sales of washers = $43,000 X = disposable income = $15,000 X 3 = Savings = $75,000 Then the forecast sales for the month of November would be: Y' = -45, (43,000) (15,000) (75,000) = -45, , , ,375 = $7, The coefficient of determination. Note that the SPSS output gives the value of R, R, and R adjusted. In this example, R = 0.99 and R = In the case of multiple regression, R is more appropriate, as was discussed previously. R = 1 (1 R n 1 ) n k = 1 - ( ) n 1 n k = (9/7) = = This identifies that 97.5 percent of total variation in sales of dryers is explained by the three explanatory variables. The remaining. percent was unexplained by the estimated equation. 3. The standard error of the estimate (S e ). This is a measure of dispersion of actual sales around the estimated equation. The output shows S e = Decision Making With 1-9 Statistics and Forecasting

32 4. Computed t. The t-statistic output is t-statistic X X X All t values are greater than a rule-of-thumb table t value of.0. (With n k 1 = = 6 degrees of freedom and a level of significance is 0.005, it can be seen from Table 1-3 that the t value is ) For a two-sided test, the level of significance was.005. By these methods, it can be concluded that all three explanatory variables selected were statistically significant. 5. F-test. From the output, we see that (Y Y) / k F = = (Y Y ) /(n k 1) Explained variation/k Unexplained variation/(n = k 1) = 9.703/0.08 = (which is given in the printout.) / / 6 At a significance level of 0.01, the F-value is far above the value of 9.78 (which is from Table 1-4). Therefore, the regression as a whole is highly significant. 6. Conclusion. Based on statistical considerations, it can be concluded that: The estimated equation had a good fit. All three variables are significant explanatory variables. The regression is highly significant. The model developed can be used as a forecasting equation with a great degree of confidence. Decision Making With 1-30 Statistics and Forecasting

33 TABLE 1- Decision Making With 1-31 Statistics and Forecasting

34 Decision Making With 1-3 Statistics and Forecasting

35 TABLE 1-3 Decision Making With 1-33 Statistics and Forecasting

36 TABLE 1-4 Decision Making With 1-34 Statistics and Forecasting

37 Decision Making With 1-35 Statistics and Forecasting

38 CHAPTER 1 REVIEW QUESTIONS The following questions are designed to ensure that you have a complete understanding of the information presented in the chapter. They do not need to be submitted in order to receive CPE credit. They are included as an additional tool to enhance your learning experience. We recommend that you answer each review question and then compare your response to the suggested solution before answering the final exam questions related to this chapter. 1. Which one of the following is a sales forecasting technique: a) linear programming (LP) b) moving average c) queuing theory d) economic order quantity (EOQ). The moving-average method of forecasting: a) is a cross-sectional forecasting method b) includes each new observation in the average as it becomes available and discards the oldest observation c) regresses the variable of interest on a related variable to develop a forecast d) derives final forecasts by adjusting the initial forecast based on the smoothing constant 3. As part of a risk analysis, a manager wishes to forecast the percentage growth in next month s sales for a particular plant using the past 30 month s sales results. Significant changes in the organization affecting sales volumes were made within the last 9 months. The most effective analysis technique to use would be: a) unweighted moving average b) queuing theory c) exponential smoothing d) linear regression analysis 4. A regression equation: a) estimates the dependent variables b) encompasses factors outside the relevant range c) is based on objective and constraint functions d) estimates the independent variable 5. Regression estimation programs employ many tools for problem definition and analysis. A scatter diagram is one of these tools. The objective of a scatter diagram is to: a) demonstrate correlations that make sure a linear (straight-line) relationship existed between Y and X in the past sample b) show frequency distribution in graphic form c) divide a universe of data into homogeneous groups d) show the vital trend and separate trivial items Decision Making With 1-36 Statistics and Forecasting

39 6. Correlation is a term frequently used in conjunction with regression analysis and is measured by the value of the coefficient of correlation, R. The best explanation of the value R is that it: a) interprets variances in terms of the independent variable b) is a measure of the relative relationship between two variables c) ranges in size from negative infinity to positive infinity d) is positive only for downward-sloping regression lines 7. In regression analysis, the coefficient of determination is a measure of: a) the amount of variation in the dependent variable explained by the independent variables b) the amount of variation in the dependent variable unexplained by the independent variables c) the slope of the regression line d) the predicted value of the dependent variable 8. In a simple linear regression model, the standard error of the estimate of Y represents: a) a range of values constructed from the regression equation results for a specified level of probability b) a variability about the least squares line that is uniform for all values of the independent variable in the sample c) a measure of variability of the actual observations from the regression line d) the proportion of the variance explained by the independent variable 9. Multicollinearity occurs when: a) a proportion of the variance is explained by the independent variable b) independent variables are correlated with each other c) observations are not independent d) a random sample fails to represent the population 10. Autocorrelation or serial correlation: a) means that observations are not independent b) defines the proportion of the variance explained by the independent variable c) means that independent variables are correlated with each other d) is the failure of random samples to represent the population 11. Two measures are commonly used for summarizing historical errors: the mean absolute deviation (MAD) and the mean. a) true b) false Decision Making With 1-37 Statistics and Forecasting

40 CHAPTER 1 SOLUTIONS AND SUGGESTED RESPONSES 1. A: Incorrect. LP is a method of minimizing or maximizing a function given certain constraints. B: Correct. With the moving average, simply take the most recent observations (n) to calculate an average. Then, use this as the forecast for the next period. Moving averages are updated as new data are received. C: Incorrect. Queuing theory is a method of determining the appropriate number of service stations (such as teller windows or cash registers) to minimize the sum of service and waiting costs. D: Incorrect. EOQ attempts to determine the order quantity that results in the lowest ordering and carrying costs. (See page 1-5 of the course material.). A: Incorrect. Cross-sectional regression analysis examines relationships among large amounts of data (e.g., many or different production methods or locations) at a particular moment in time. B: Correct. Moving averages are averages that are updated as new information is received. With the moving average, a manager simply employs the most recent observations to calculate an average, which is used as the forecast for the next period. C: Incorrect. Regression analysis relates the forecast to changes in particular variables. D: Incorrect. Under exponential smoothing, each forecast equals the sum of the last observation times the smoothing constant, plus the last forecast times one minus the constant. (See page 1-5 of the course material.) 3. A: Incorrect. An unweighted average will not give more importance to more recent data. B: Incorrect. Queuing theory is used to minimize the cost of waiting lines. C: Correct. Exponential smoothing is a popular technique for short-run forecasting by financial managers. It uses a weighted average of past data as the basis for a forecast. The formula for exponential smoothing is: Y' t+1 = αy t + (1 α)y' t or Y' new = αy old + (1 α)y' old where Y' new = Exponentially smoothed average to be used as the forecast. Y old = Most recent actual data. Y' old = Most recent smoothed forecast. α = Smoothing constant. The higher the α, the higher the weight given to the more recent information. D: Incorrect. Linear regression analysis determines the equation for the relationship among variables. It does not give more importance to more recent data. (See page 1-7 of the course material.) Decision Making With 1-38 Statistics and Forecasting

41 4. A: Correct. Regression analysis is a statistical procedure for estimating mathematically the average relationship between the dependent variable and the independent variable(s). For example, regression analysis is used to estimate a dependent variable (such as cash collections from customers) given a known independent variable (such as credit sales). B: Incorrect. Regression results are limited to observations within the relevant range. C: Incorrect. Regression analysis does not use constraint functions. D: Incorrect. The dependent variable is estimated using regression analysis. (See page 1-14 of the course material.) 5. A: Correct. The objective of a scatter diagram is to demonstrate correlations. Each observation is represented by a dot on a graph corresponding to a particular value of X (the independent variable) and Y (the dependent variable). Also, make sure that a linear (straight-line) relationship existed between Y and X in the past sample. B: Incorrect. The objective of a histogram is to show frequency distribution in graphic form. C: Incorrect. The objective of stratification is to divide a universe of data into homogeneous groups. D: Incorrect. Regression analysis is used to find trend lines. (See page 1-15 of the course material.) 6. A: Incorrect. The coefficient of correlation (R) relates the two variables to each other. B: Correct. The coefficient of correlation (R) measures the strength of the linear relationship between the dependent and independent variables. The magnitude of R is independent of the scales of measurement of x and y. The coefficient lies between 1.0 and A value of zero indicates no relationship between the x and y variables. A value of +1.0 indicates a perfectly direct relationship, and a value of 1.0 indicates a perfectly inverse relationship. C: Incorrect. The coefficient of correlation (R) lies between 1.0 and D: Incorrect. A downward-sloping regression line indicates a negative correlation. A downward slope means that y decreases as x increases. (See pages 1-19 to 1-0 of the course material.) Decision Making With 1-39 Statistics and Forecasting

42 7. A: Correct. Squaring the coefficient of correlation gives the coefficient of determination, which is a measure of the amount of variation in a dependent variable that can be explained by independent variables. B: Incorrect. The complement of the coefficient of determination is the unexplained variation. C: Incorrect. The slope is the change in the dependent variable in relation to the change in independent variables. D: Incorrect. The predicted value of the dependent variable is calculated by the regression formula (Y = a + bx for simple regression). (See page 1-0 of the course material.) 8. A: Incorrect. It describes a confidence interval. B: Incorrect. It describes constant variance. C: Correct. The standard error of the estimate represents the variance of actual observations from the regression line. It is computed as: (Y - Y' ) n - This statistic can be used to gain some idea of the accuracy of a prediction. D: Incorrect. It describes the coefficient of determination. (See page 1-1 of the course material.) 9. A: Incorrect. The coefficient of determination is a measure of the amount of variation in a dependent variable that can be explained by independent variables. B: Correct. Multicollinearity occurs when independent variables are correlated and interfere with each other. C: Incorrect. Autocorrelation is a problem often encountered in regression analysis. Autocorrelation usually indicates that an important part of the variation of the dependent variable has not been explained. The best solution to this problem is to search for other explanatory variables to include in the regression equation. D: Incorrect. Bias occurs when a random sample parameter fails to represent the population parameters (for example, mean). (See page 1-5 of the course material.) Decision Making With 1-40 Statistics and Forecasting

43 10. A: Correct. Autocorrelation and serial correlation are synonyms meaning that the observations are not independent. For example, certain costs may rise with an increase in volume but not decline with a decrease in volume. B: Incorrect. This is the definition of the coefficient of determination. C: Incorrect. This is the definition of multicollinearity. D: Incorrect. This is the definition of bias. (See page 1-5 of the course material.) 11. A: True is incorrect. Two measures are commonly used for summarizing historical errors: the mean absolute deviation (MAD) and the mean squared error (MSE). B: False is correct. Mean is not used for forecasting accuracy. Mean is a measure of central tendency. (See page 1-6 of the course material.) Decision Making With 1-41 Statistics and Forecasting

44 Chapter : Making Use of Quantitative Decision Making Upon completion of this chapter, you will be able to: Define decision making under conflict. Demonstrate how a zero sum game works. Construct a decision tree. List applications of linear programming. Explain how learning curves affect labor cost structure. Decide when and how much to order. Discuss the purpose of queuing theory. Quantitative methods (or models) are used in operations research or management science. They refer to sophisticated mathematical and statistical techniques for solving problems pertaining to managerial planning and decision making. Numerous such techniques are available, some of which are discussed in this chapter. They are: Decision making under certainty and conflict Decision making under uncertain conditions Decision theory Linear programming and shadow prices Learning curve Inventory planning and control Queuing models I. Decision Making Under Risk, Certainty, and Conflict Decision making under risk involves decisions made when the probability of occurrence of the different states of nature is known. Decision making involves managing three major elements: Decision strategy. A decision maker implements a decision strategy which utilizes known existing organizational resources. States of nature. Elements of the environment over which the manager has little or no control. States of nature include the weather, political environment, economy, technological developments, etc. They can dramatically affect the outcome of any decision strategy. Outcome. The result of the interaction of the implementation of a decision strategy with the states of nature. Because of the variability of the states of nature, outcomes can be extremely difficult to forecast. The outcomes of a decision strategy O (the dependent variable) is a function of the interaction of the two independent variables, D (decision strategies) and S (the states of nature). Figure -1 shows a decision matrix, which is an approach to decision making under risk. Decision matrix, also called payoff table, is a matrix consisting of the decision alternatives, the states of nature, and the decision outcomes. The rows are strategic choices a manager can make while the columns represent decision outcomes. An outcome O ij is a function of a decision strategy D i and state of nature S j. Making Use of Quantitative Decision Making -1

45 FIGURE -1 DECISION MATRIX States of Nature Strategies S 1 S S 3.. S n D 1 O 11 O 1 * * * O 1n D O 1 O * * * O n D 3 * * *. * * *. D m O i1 O i * * * O mn Mathematically this relationship can be expressed as: O ij = A. DECISION MAKING UNDER CERTAINTY f(d S ) where i = 1,, m and j = 1,, n. i j This is the simplest type of decision making since it has a known state of nature. Therefore, the outcomes are the direct result of the chosen decision strategy and can be predicted with certainty. In reality, this situation rarely occurs. The manager simply evaluates all the available decision strategies and then chooses the one best meeting the outcome criteria. Various optimization techniques can be utilized to maximize a decision strategy. Decision making under certainty occurs with problems that can be analyzed using basic inventory models, break-even analysis, linear programming, incremental analysis, and other methods where an outcome having one state of nature can be determined. B. DECISION MAKING UNDER CONFLICT In this situation, the decision maker is opposed by another party who is designing states of nature strategies to offset the decision maker s strategy to gain a competitive advantage. The decision maker must develop decision strategies to defeat an opponent s state of nature control strategy. This is the ideal setting for developing game strategies. Games are dependent on rules governing a competitive situation where the number of players, strategies, states of nature between the players, and degree of conflict control the outcomes. C. THE TYPES OF GAMES Games are classified according to the degree of conflict of interest between the opponents. A zero sum game has a perfect inverse relationship between the gains and losses of the opponents. One opponent s gain is the other s loss. The total sum remains the same. Making Use of Quantitative Decision Making -

46 In nonzero-sum games, the gains of one participant do not necessarily represent a comparable loss for the other party to the game. In the business environment, most competitive situations are nonzero-sum games. D. ZERO SUM GAME In a zero sum game, the gains and losses are always equal. No player can gain more than the other player loses. Therefore, the game is always in equilibrium. The simplest type of zero sum game is the two-person zero sum game. Each player has a choice of game strategies. Since each player s gain will equal the other s loss, the outcome for each game strategy is known to each player. In the two-person zero sum game, the outcomes can be expressed numerically. A twoperson zero sum game outcome matrix is show in Example 1. A positive number indicates a payoff to the player for rows A, and a negative number indicates a payoff to the player for columns B. In Example 1, column Strategy F, the maximum any player can win/lose is 11. Example 1 Two-player zero sum game outcome matrix E. PURE STRATEGY A pure strategy exists when there is one strategy for player A and one for player B that will be played every time. An equilibrium point is reached when it is at an optimum point for each respective player. This is termed a saddle point. A saddle point occurs where it is both the smallest numerical value in its row and the largest numerical value in its column. Example Player B Player A Strategy F Strategy H r 1 r Strategy D p 1 A wins 5 A wins 8 Strategy E p A wins 6 B wins (or A loses ) Example shows a sample saddle point in a two-person zero sum game. Since 13 is the row minimum and the column maximum, it is the saddle point strategy, the value of this game is 13 where the first choice Player A gains 13 while Player B loses 13. Saddle Point for two-person Zero Sum Game Player B Player A Making Use of Quantitative Decision Making -3

47 F. MIXED STRATEGY When no player has one strategy that will be used each time, then there is no pure strategy used in the zero sum game. In this case the optimum point, or saddle point, is found using a mixed strategy. In this case, each player s strategy is chosen using a random number process. Nonetheless, one player s gain is another player s loss. Example 3 Using the Two-Player Zero Sum Game Outcome Matrix in Example 1, it is possible to establish a mixed strategy. Assuming p 1 and p are the probabilities for A s strategies, and n 1 and n are the probabilities for B s strategies, their values can be determined using the following process in section II. II. Decision Making Under Uncertain Conditions A. CONDITIONS FOR DECISION MAKING Decisions are made under certainty or under uncertainty (or risk). Under certainty implies that there is only one event and therefore only one outcome for each action. Under uncertainty, which is more common, several events are involved for each action and with each a different probability of occurrence. Under uncertainty, it is often helpful to compute the following: Expected value Standard deviation Coefficient of variation B. EXPECTED VALUE For decisions involving uncertainty, the concept of expected value ( r ) provides a rational means for selecting the best course of action. Expected value is defined as a weighted average using the probabilities as weights. It is found by multiplying the probability of each outcome by its payoff: r = r i p i where r i is the outcome for the ith possible event and p i is the probability of occurrence of that outcome. Note: A rational economic decision maker (one completely guided by objective criteria) will use expected monetary value to maximize gains under conditions of uncertainty because (s)he is risk-neutral. Expected value represents the long-term average payoff for repeated trials. The best choice is the one having the highest expected value (sum of the products of the possible outcomes and their respective probabilities). C. STANDARD DEVIATION Whenever we talk about the expected value, one statistic that goes with it is standard deviation. Standard deviation is a statistic that measures the tendency of data to be spread out or is also a measure of the dispersion of a probability distribution. The Making Use of Quantitative Decision Making -4

48 smaller the deviation, the tighter the distribution and thus the lower the riskiness of the project. It is intuitively a margin of error associated with a given expected value. MBAs can make important inferences from past data with this measure. It is the square root of the mean of the squared deviations from the expected value (r ). The standard deviation, denoted with the Greek letter σ (read as sigma), is calculated as follows: To calculate σ, follow these steps: σ = (ri r) pi Step 1. First compute the expected rate of return (r ) Step. Subtract each possible return from r to obtain a set of deviations (r i - r ) Step 3. Square each deviation, multiply the squared deviation by the probability of occurrence for its respective return, and sum these products to obtain the variance ( σ ): σ = (ri r) pi Step 4. Finally, take the square root of the variance to obtain the standard deviation (σ). The standard deviation can be used to measure the variation of such items as the expected profits, expected contribution margin, or expected cash flows. It can also be used to assess the absolute risk associated with investment projects. Tip: The higher the standard deviation, the higher the risk. D. COEFFICIENT OF VARIATION The coefficient of variation is a popular measure of relative dispersion, or relative risk. It represents the degree of risk per unit of return. It is computed by dividing the standard deviation by the expected value: Example 4 σ/r Consider two investment proposals, A and B, with the following probability distribution of cash flows in each of the next five years: Cash Inflows Probability (.) (.6) (.) Project A $ Project B $ The expected value of the cash inflow is computed as follows: Project A $00(.) + 300(.6) + 400(.) = $300 Project B $100(.) + 300(.6) + 500(.) = $300 Making Use of Quantitative Decision Making -5

49 The standard deviations are computed as follows: A = For A: σ = ($00 300) (.) + ( ) (.6) + ( ) (.) $63. 5 B = For B: σ = ($ ) (.) + ( ) (.6) + ( ) (.) $ The coefficients of variation are computed as follows: For A: $63.5/$300 =.1 For B: $16.49/$300 =.4 Conclusions: Proposal B is more risky than proposal A since its standard deviation is greater. Because the coefficient is a relative measure of risk, the degree of risk is also greater for Project B. III. Decision Theory A. DEFINING DECISION THEORY Decision theory refers to a systematic approach to making decisions, particularly under conditions of uncertainty. While statistics such as expected value and standard deviation are essential for making the best choice, the decision problem can best be approached by using decision theory. Decision theory utilizes an organized approach, such as a decision matrix (or payoff table). It is characterized by: The row. Each row represents a set of available alternative courses of action. The column. Each column represents the state of nature, or conditions that are likely to occur and over which there is no control. The entries. These appear in the body of the table and represent the outcome of the decision, known as payoffs. These may be in the form of costs, revenues, profits, contribution margins or cash flows. By computing expected value of each action, it is possible to pick the best one. Example 5 Assume the following probability distribution of daily demand for strawberries: Daily demand Probability Also assume that unit cost = $3, selling price = $5 (i.e., profit on sold unit = $), and salvage value on unsold units = $ (i.e., loss on unsold unit = $1). The company can stock either 0, 1,, or 3 units. The problem is: How many units should be stocked each day? Assume that units from one day cannot be sold the next day. The payoff table can be constructed as follows: Making Use of Quantitative Decision Making -6

50 State of Nature Demand Expected value Stock (probability) (.) (.3) (.3) (.) 0 $ $0 Actions * ** *Profit for (stock, demand 1) equals (no. of units sold)(profit per unit) (no. of units unsold)(loss per unit) = (1)($5-3) - (1)($3 - ) = $ - $1 = $1 **Expected value for (stock ) is: -(.) + 1(.3) + 4 (.3) + 4 (.) = $1.90 The optimal stock action is the one with the highest expected monetary value (stock units). B. THE ROLE OF PERFECT INFORMATION IN DECISION THEORY Suppose it is possible to obtain a perfect prediction of which event (state of nature) will occur. The expected value with perfect information would be the total expected value of actions selected on the assumption of a perfect forecast. The expected value of perfect information (EVPI) can then be computed as: EVPI = Expected value with perfect information minus the expected value with existing information. Example 6 From the payoff table in Example 5, with perfect information, one can make the following analysis: Alternatively, State of Nature Demand Expected value Stock (probability) (.) (.3) (.3) (.) 0 $0 $0 Actions $3.00 $0 (.) + (.3) + 4 (.3) + 6 (.) = $3.00 Conclusions: The optimal stock action is stock, with the highest expected value of $1.90. Thus, with existing information, the best course of action is to select stock units to obtain $1.90. With perfect information (forecast), it is possible to make as much as $3. Therefore, the expected value of perfect information (EVPI) = $ $1.90, or $1.10. This is the maximum price one should be willing to pay for additional information. Making Use of Quantitative Decision Making -7

51 C. USING A DECISION TREE A decision tree is another approach used in discussions of decision making under uncertainty. It is a pictorial representation of a decision situation. As in the case of the decision matrix approach, it shows decision alternatives, states of nature, probabilities attached to the state of nature, and conditional benefits and losses. The decision tree approach is most useful in a sequential decision situation. Example 7 Assume XYZ Corporation wishes to introduce one of two products to the market this year. The probabilities and present values (PV) of projected cash inflows are given below: Product Initial investment PV of cash inflows Probabilities A $5, $450, , , B 80, , , , A decision tree analyzing the two products is given in Figure -. FIGURE - Making Use of Quantitative Decision Making -8

52 Based on the expected NPV, choose product A over product B. Note: This analysis fails to recognize the risk factor in project analysis. IV. Linear Programming and Shadow Prices A. DEFINING LINEAR PROGRAMMING Linear programming (LP) is a mathematical technique designed to determine an optimal decision (or an optimal plan) chosen from a large number of possible decisions. The optimal decision is the one that meets the specified objective of the company, subject to various restrictions or constraints. It concerns itself with the problem of allocating scarce resources among competing activities in an optimal manner. The optimal decision yields the highest profit, contribution margin (CM) or revenue, or the lowest cost. B. THE COMPONENTS OF LINEAR PROGRAMMING A linear programming model consists of two important ingredients: 1. Objective function. The company must define the specific objective to be achieved.. Constraints. Constraints are in the form of restrictions on availability of resources or meeting minimum requirements. As the name linear programming indicates, both the objective function and constraints must be in linear form. Example 8 A firm wishes to find an optimal product mix. The optimal mix would be the one that maximizes its total profit or contribution margin (CM) within the allowed budget and production capacity. Or the firm may want to determine a least cost combination of input materials while meeting production requirements, employing production capacities, and using available employees. C. THE APPLICATIONS OF LINEAR PROGRAMMING Applications of LP are numerous. They include: Selecting least-cost mix of ingredients for manufactured products Developing an optimal budget Determining an optimal investment portfolio (or asset allocation) Allocating an advertising budget to a variety of media. Scheduling jobs to machines Determining a least-cost shipping pattern Scheduling flights Gasoline blending Optimal manpower allocation Selecting the best warehouse location to minimize shipping costs. D. FORMULATION OF LP To formulate an LP problem, follow these steps: 1. Define the decision variables that must be solved for.. Express the objective function and constraints in terms of these decision variables. Note: All the expressions must be in linear form. Making Use of Quantitative Decision Making -9

53 In the following example, this technique is used to find the optimal product mix. Example 9 The Omni Furniture Manufacturing Company produces two products: desk and table. Both products require time in two processing departments, the Assembly Department and the Finishing Department. Data on the two products are as follows: Products Available Processing Desk Table Hours Assembly hours Finishing 3 90 Contribution Margin Per Unit $5 $40 The company wants to find the most profitable mix of these two products. Step 1: Define the decision variables as follows: A = Number of units of desk to be produced B = Number of units of table to be produced Step : The objective function to maximize total contribution margin (CM) is expressed as: Total CM = 5A + 40B Then, formulate the constraints as inequalities: A + 4B < 100 (Assembly constraint) 3A + B < 90 (Finishing constraint) In addition, implicit in any LP formulation are the constraints that restrict A and B to be nonnegative, i.e., The LP model is: E. SOLVING LP PROBLEMS A, B > 0 Maximize: Total CM = 5A + 40B Subject to: A + 4B < 100 3A + B < 90 A, B > 0 There are several methods available to solve LP problems. Two common ones are: The simplex method. This is the most commonly used method of solving LP problems. It uses an algorithm, which is an iteration method of computation, to move from one solution to another until it reaches the best one. The graphical method. This solution is easier to use but limited to the LP problems involving two (or at most three) decision variables. Making Use of Quantitative Decision Making -10

54 To use the graphical method, follows these five steps: Step 1: Change inequalities to equalities. Step : Graph the equalities. To graph the equality, set one variable equal to zero and find the value of the other. Then, connect those two points on the graph and mark these intersections on the axes, connecting them with a straight line. Step 3: Identify the correct side for the original inequalities by shading. Repeat steps 1-3 for each constraint. Step 4: After all this, identify the feasible region, the area of feasible solutions. Step 5: Solve the constraints (expressed as equalities) simultaneously for the various corner points of the feasible region. Determine the profit or contribution margin at all corners in the region. Example 10 Using the data and the LP model from Example 9, obtain the feasible region (shaded area) by going through steps one through four. Then evaluate all of the corner points as follows: Corner Points Contribution Margin A B $5A + $40B (a) 30 0 $5(30) + $40(0) = $750 (b) (0) + 40(15) = 1,100 (c) 0 5 5(0) + (40)(5) = 1,000 (d) 0 0 5(0) + 40(0) = 0 Conclusion: The corner 0A, 15B produces the most profitable solution. (See Figure - 3.) FIGURE -3 THE FEASIBLE REGION AND CORNER POINTS Making Use of Quantitative Decision Making -11

55 F. SHADOW PRICES (OPPORTUNITY COSTS) After having solved an LP problem, it might still be desirable to know whether it pays to add capacity in hours in a particular department. For example, what monetary value would be gained by adding an hour per week of assembly time? This monetary value is usually the additional contribution that could be earned. This amount is the shadow price of a given resource. Shadow prices constitute a form of opportunity cost that is considered as the contribution margin that would be lost by not adding capacity. To justify a decision in favor of a short-term capacity increase, be sure that the shadow price exceeds the actual price of that expansion. For example, suppose that the shadow price of an hour of assembly capacity is $6.50 while the actual market price is $8.00. That means it does not pay to obtain an additional hour of the assembly capacity. Here is how to compute shadow prices (or opportunity cost): 1. Add one hour (preferably more than one hour to make it easier to show graphically) to the constraint of a given LP problem under consideration.. Resolve the problem and find the maximum CM. 3. Compute the difference between the CM of the original LP problem and the CM determined in step, which is the shadow price. Example 11 Using the data in Example 10, calculate the shadow price of the assembly capacity. To make it easier to show graphically, add 8 hours of capacity to the assembly department, rather than one hour. The new assembly constraint and the resulting feasible region are shown in Figure -4. Then, evaluate all of the corner points in the new feasible region in terms of their CM, as follows: Corner Points Contribution Margin A B $5A + $40B (a) 30 0 $5(30) + $40(0) = $750 (b) (18) + 40(18) = 1,170 (c) 0 7 5(0) + (40)(7) = 1,080 (d) 0 0 5(0) + 40(0) = 0 The new optimal solution, Corner b (18A, 18B) has total CM of $1,170 per week. The shadow price of the assembly capacity is $70 ($1,170 - $1,100 = $70) or $8.75 per hour ($70/8 hours = $8.75). Conclusion: The company would be willing to pay up to $70 to obtain an additional 8 hours of the assembly capacity per week, or $8.75 per hour per week. In other words, the company s opportunity cost of not adding an additional hour is $8.75. Making Use of Quantitative Decision Making -1

56 FIGURE -4 THE FEASIBLE REGION AND CORNER POINTS G. HOW TO USE THE COMPUTER FOR LINEAR PROGRAMMING Computer LP software packages, such as LINDO (Linear Interactive and Discrete Optimization) ( What's Best! ( or Microsoft Excel, can be used to quickly solve an LP problem. Figure -5 shows a computer output by an LP software program for the LP model. Figure -6 presents an Excel LP solution. Note: The printout shows the following optimal solution: A = 0 units B = 15 units CM = $1,100 Shadow prices are: Assembly capacity = $8.75 Finishing capacity = $.50 Making Use of Quantitative Decision Making -13

57 FIGURE -5 THE LP COMPUTER OUTPUT Making Use of Quantitative Decision Making -14

58 FIGURE -6 THE EXCEL LP INPUT AND OUTPUT V. Learning Curve A. USING THE LEARNING CURVE TO ESTIMATE LABOR HOURS In manufacturing, labor hours are often observed to decrease in a definite pattern as labor operations are repeated. More specifically, as the cumulative production doubles, the cumulative average time required per unit will be reduced by some constant percentage, ranging typically from 10 percent to 0 percent. This reduction, and hence related costs, is referred to as the learning curve effect. B. EXPRESSING THE LEARNING CURVE RELATIONSHIP By convention, learning curves are referred to in terms of the complements of their improvement rates. For example, an 80 percent learning curve denotes a 0 percent decrease in unit time with each doubling of repetitions. Making Use of Quantitative Decision Making -15

Spreadsheets in Education (ejsie)

Spreadsheets in Education (ejsie) Spreadsheets in Education (ejsie) Volume 2, Issue 2 2005 Article 5 Forecasting with Excel: Suggestions for Managers Scott Nadler John F. Kros East Carolina University, nadlers@mail.ecu.edu East Carolina

More information

Chapter 3. Table of Contents. Introduction. Empirical Methods for Demand Analysis

Chapter 3. Table of Contents. Introduction. Empirical Methods for Demand Analysis Chapter 3 Empirical Methods for Demand Analysis Table of Contents 3.1 Elasticity 3.2 Regression Analysis 3.3 Properties & Significance of Coefficients 3.4 Regression Specification 3.5 Forecasting 3-2 Introduction

More information

Forecasting Introduction Version 1.7

Forecasting Introduction Version 1.7 Forecasting Introduction Version 1.7 Dr. Ron Tibben-Lembke Sept. 3, 2006 This introduction will cover basic forecasting methods, how to set the parameters of those methods, and how to measure forecast

More information

Forecasting Survey. How far into the future do you typically project when trying to forecast the health of your industry? less than 4 months 3%

Forecasting Survey. How far into the future do you typically project when trying to forecast the health of your industry? less than 4 months 3% Forecasting Forecasting Survey How far into the future do you typically project when trying to forecast the health of your industry? less than 4 months 3% 4-6 months 12% 7-12 months 28% > 12 months 57%

More information

Selection of a Forecasting Technique for Beverage Production: A Case Study

Selection of a Forecasting Technique for Beverage Production: A Case Study World Journal of Social Sciences Vol. 6. No. 3. September 2016. Pp. 148 159 Selection of a Forecasting Technique for Beverage Production: A Case Study Sonia Akhter**, Md. Asifur Rahman*, Md. Rayhan Parvez

More information

Multiple Regression. Dr. Tom Pierce Department of Psychology Radford University

Multiple Regression. Dr. Tom Pierce Department of Psychology Radford University Multiple Regression Dr. Tom Pierce Department of Psychology Radford University In the previous chapter we talked about regression as a technique for using a person s score on one variable to make a best

More information

COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE

COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE COORDINATING DEMAND FORECASTING AND OPERATIONAL DECISION-MAKING WITH ASYMMETRIC COSTS: THE TREND CASE ABSTRACT Robert M. Saltzman, San Francisco State University This article presents two methods for coordinating

More information

MANAGERIAL ECONOMICS WILEY A JOHN WILEY & SONS, INC., PUBLICATION. A Mathematical Approach

MANAGERIAL ECONOMICS WILEY A JOHN WILEY & SONS, INC., PUBLICATION. A Mathematical Approach MANAGERIAL ECONOMICS A Mathematical Approach M. J. ALHABEEB L. JOE MOFFITT Isenberg School of Management University of Massachusetts Amherst, MA, USA WILEY A JOHN WILEY & SONS, INC., PUBLICATION PREFACE

More information

Descriptive Statistics Tutorial

Descriptive Statistics Tutorial Descriptive Statistics Tutorial Measures of central tendency Mean, Median, and Mode Statistics is an important aspect of most fields of science and toxicology is certainly no exception. The rationale behind

More information

OF EQUIPMENT LEASE FINANCING VOLUME 33 NUMBER 3 FALL Equipment Finance Market Forecasting By Blake Reuter

OF EQUIPMENT LEASE FINANCING VOLUME 33 NUMBER 3 FALL Equipment Finance Market Forecasting By Blake Reuter Articles in the Journal of Equipment Lease Financing are intended to offer responsible, timely, in-depth analysis of market segments, finance sourcing, marketing and sales opportunities, liability management,

More information

Regression Analysis I & II

Regression Analysis I & II Data for this session is available in Data Regression I & II Regression Analysis I & II Quantitative Methods for Business Skander Esseghaier 1 In this session, you will learn: How to read and interpret

More information

Introduction to Analytics Tools Data Models Problem solving with analytics

Introduction to Analytics Tools Data Models Problem solving with analytics Introduction to Analytics Tools Data Models Problem solving with analytics Analytics is the use of: data, information technology, statistical analysis, quantitative methods, and mathematical or computer-based

More information

Financial Forecasting:

Financial Forecasting: Financial Forecasting: Tools and Applications Course Description Business and financial forecasting is of extreme importance to managers at practically all levels. It is required for top managers to make

More information

Business Math Curriculum Guide Scranton School District Scranton, PA

Business Math Curriculum Guide Scranton School District Scranton, PA Business Math Scranton School District Scranton, PA Business Math Prerequisite : Geometry 11 or Applied Geometry 11 Course Description: This course is dedicated to real world applications of Algebra Concepts.

More information

WINDOWS, MINITAB, AND INFERENCE

WINDOWS, MINITAB, AND INFERENCE DEPARTMENT OF POLITICAL SCIENCE AND INTERNATIONAL RELATIONS Posc/Uapp 816 WINDOWS, MINITAB, AND INFERENCE I. AGENDA: A. An example with a simple (but) real data set to illustrate 1. Windows 2. The importance

More information

SECTION 11 ACUTE TOXICITY DATA ANALYSIS

SECTION 11 ACUTE TOXICITY DATA ANALYSIS SECTION 11 ACUTE TOXICITY DATA ANALYSIS 11.1 INTRODUCTION 11.1.1 The objective of acute toxicity tests with effluents and receiving waters is to identify discharges of toxic effluents in acutely toxic

More information

Chapter 5 Demand Forecasting

Chapter 5 Demand Forecasting Chapter 5 Demand Forecasting TRUE/FALSE 1. One of the goals of an effective CPFR system is to minimize the negative impacts of the bullwhip effect on supply chains. 2. The modern day business environment

More information

How to Get More Value from Your Survey Data

How to Get More Value from Your Survey Data Technical report How to Get More Value from Your Survey Data Discover four advanced analysis techniques that make survey research more effective Table of contents Introduction..............................................................3

More information

A Forecasting Case for TeleTech Customer Care Management Phils., Inc.

A Forecasting Case for TeleTech Customer Care Management Phils., Inc. A Forecasting Case for TeleTech Customer Care Management Phils., Inc. Stephan Earl S. Chow Decision Sciences & Innovation Department De La Salle University 2401 Taft Avenue, Manila, Philippines 1004 Email:

More information

Basic Statistics, Sampling Error, and Confidence Intervals

Basic Statistics, Sampling Error, and Confidence Intervals 02-Warner-45165.qxd 8/13/2007 5:00 PM Page 41 CHAPTER 2 Introduction to SPSS Basic Statistics, Sampling Error, and Confidence Intervals 2.1 Introduction We will begin by examining the distribution of scores

More information

STATISTICS PART Instructor: Dr. Samir Safi Name:

STATISTICS PART Instructor: Dr. Samir Safi Name: STATISTICS PART Instructor: Dr. Samir Safi Name: ID Number: Question #1: (20 Points) For each of the situations described below, state the sample(s) type the statistical technique that you believe is the

More information

GSU College of Business MBA Core Course Learning Outcomes. Updated 9/24/2015

GSU College of Business MBA Core Course Learning Outcomes. Updated 9/24/2015 GSU College of Business MBA Core Course Learning Outcomes Updated 9/24/2015 ACCT 6100 ACCT 7101 ECON 6100 ECON 7500 FIN 7101 MIS 7101 MGMT 6100 MGMT 6700 MGMT 7400 MGMT 7500 MGMT 7600 MGMT 8900 MKTG 7100

More information

STATISTICAL TECHNIQUES. Data Analysis and Modelling

STATISTICAL TECHNIQUES. Data Analysis and Modelling STATISTICAL TECHNIQUES Data Analysis and Modelling DATA ANALYSIS & MODELLING Data collection and presentation Many of us probably some of the methods involved in collecting raw data. Once the data has

More information

ENGG1811: Data Analysis using Excel 1

ENGG1811: Data Analysis using Excel 1 ENGG1811 Computing for Engineers Data Analysis using Excel (weeks 2 and 3) Data Analysis Histogram Descriptive Statistics Correlation Solving Equations Matrix Calculations Finding Optimum Solutions Financial

More information

Lesson-9. Elasticity of Supply and Demand

Lesson-9. Elasticity of Supply and Demand Lesson-9 Elasticity of Supply and Demand Price Elasticity Businesses know that they face demand curves, but rarely do they know what these curves look like. Yet sometimes a business needs to have a good

More information

CHAPTER 3. Quantitative Demand Analysis

CHAPTER 3. Quantitative Demand Analysis CHAPTER 3 Quantitative Demand Analysis Copyright 2014 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. Chapter Outline

More information

WRITTEN PRELIMINARY Ph.D. EXAMINATION. Department of Applied Economics. University of Minnesota. June 16, 2014 MANAGERIAL, FINANCIAL, MARKETING

WRITTEN PRELIMINARY Ph.D. EXAMINATION. Department of Applied Economics. University of Minnesota. June 16, 2014 MANAGERIAL, FINANCIAL, MARKETING WRITTEN PRELIMINARY Ph.D. EXAMINATION Department of Applied Economics University of Minnesota June 16, 2014 MANAGERIAL, FINANCIAL, MARKETING AND PRODUCTION ECONOMICS FIELD Instructions: Write your code

More information

Managerial Economics

Managerial Economics Managerial Economics Estimating Demand Functions Rudolf Winter-Ebmer Johannes Kepler University Linz Winter Term 2014 Winter-Ebmer, Managerial Economics: Unit 2 - Demand Estimation 1 / 21 Why do you need

More information

Managerial Decision-Making Introduction To Using Excel In Forecasting

Managerial Decision-Making Introduction To Using Excel In Forecasting Managerial Decision-Making Introduction To Using Excel In May 28-31, 2012 Thomas H. Payne, Ph.D. Dunagan Chair of Excellence in Banking Chair, Department of Accounting, Finance, Economics and Political

More information

Chapter 11. Decision Making and Relevant Information Linear Programming as a Decision Facilitating Tool

Chapter 11. Decision Making and Relevant Information Linear Programming as a Decision Facilitating Tool Chapter 11 Decision Making and Relevant Information Linear Programming as a Decision Facilitating Tool 1 Introduction This chapter explores cost accounting for decision facilitating purposes It focuses

More information

Business Intelligence, 4e (Sharda/Delen/Turban) Chapter 2 Descriptive Analytics I: Nature of Data, Statistical Modeling, and Visualization

Business Intelligence, 4e (Sharda/Delen/Turban) Chapter 2 Descriptive Analytics I: Nature of Data, Statistical Modeling, and Visualization Business Intelligence, 4e (Sharda/Delen/Turban) Chapter 2 Descriptive Analytics I: Nature of Data, Statistical Modeling, and Visualization 1) One of SiriusXM's challenges was tracking potential customers

More information

AN EXPERIMENTAL ANALYSIS OF ADVERTISING STRATEGIES

AN EXPERIMENTAL ANALYSIS OF ADVERTISING STRATEGIES AN EXPERIMENTAL ANALYSIS OF ADVERTISING STRATEGIES AND ADVERTISING FUNCTIONS Kenneth Goosen Micro Business Publications, Inc. krgoosen@cei.net ABSTRACT This paper explores the relationship of different

More information

EnterpriseOne JDE5 Forecasting PeopleBook

EnterpriseOne JDE5 Forecasting PeopleBook EnterpriseOne JDE5 Forecasting PeopleBook May 2002 EnterpriseOne JDE5 Forecasting PeopleBook SKU JDE5EFC0502 Copyright 2003 PeopleSoft, Inc. All rights reserved. All material contained in this documentation

More information

23 Perfect Competition

23 Perfect Competition 23 Perfect Competition Learning Objectives After you have studied this chapter, you should be able to 1. define price taker, total revenues, marginal revenue, short-run shutdown price, short-run breakeven

More information

TDWI strives to provide course books that are contentrich and that serve as useful reference documents after a class has ended.

TDWI strives to provide course books that are contentrich and that serve as useful reference documents after a class has ended. Previews of TDWI course books offer an opportunity to see the quality of our material and help you to select the courses that best fit your needs. The previews cannot be printed. TDWI strives to provide

More information

Building a Mathematical Model for Predicting the Cost of the Communication Towers Projects Using Multifactor Linear Regression Technique

Building a Mathematical Model for Predicting the Cost of the Communication Towers Projects Using Multifactor Linear Regression Technique International Journal of Construction Engineering and Management 2016, 5(1): 25-29 DOI: 10.5923/j.ijcem.20160501.03 Building a Mathematical Model for Predicting the Cost of the Communication Towers Projects

More information

1) Operating costs, such as fuel and labour. 2) Maintenance costs, such as overhaul of engines and spraying.

1) Operating costs, such as fuel and labour. 2) Maintenance costs, such as overhaul of engines and spraying. NUMBER ONE QUESTIONS Boni Wahome, a financial analyst at Green City Bus Company Ltd. is examining the behavior of the company s monthly transportation costs for budgeting purposes. The transportation costs

More information

The Dummy s Guide to Data Analysis Using SPSS

The Dummy s Guide to Data Analysis Using SPSS The Dummy s Guide to Data Analysis Using SPSS Univariate Statistics Scripps College Amy Gamble April, 2001 Amy Gamble 4/30/01 All Rights Rerserved Table of Contents PAGE Creating a Data File...3 1. Creating

More information

a) Measures are unambiguous quantities, whereas indicators are devised from common sense understandings

a) Measures are unambiguous quantities, whereas indicators are devised from common sense understandings Chapter-1: QUANTITATIVE TECHNIQUES Self Assessment Questions 1. An operational definition is: a) One that bears no relation to the underlying concept b) An abstract, theoretical definition of a concept

More information

ENGINEERING ECONOMICS AND FINANCIAL ACCOUNTING 2 MARKS

ENGINEERING ECONOMICS AND FINANCIAL ACCOUNTING 2 MARKS ENGINEERING ECONOMICS AND FINANCIAL ACCOUNTING 2 MARKS 1. What is managerial economics? It is the integration of economic theory with business practice for the purpose of facilitating decision making and

More information

Chapter 2 Maintenance Strategic and Capacity Planning

Chapter 2 Maintenance Strategic and Capacity Planning Chapter 2 Maintenance Strategic and Capacity Planning 2.1 Introduction Planning is one of the major and important functions for effective management. It helps in achieving goals and objectives in the most

More information

Chapter 1 INTRODUCTION TO SIMULATION

Chapter 1 INTRODUCTION TO SIMULATION Chapter 1 INTRODUCTION TO SIMULATION Many problems addressed by current analysts have such a broad scope or are so complicated that they resist a purely analytical model and solution. One technique for

More information

EXAMINERS REPORT ON THE PERFORMANCE OF CANDIDATES CSEE, 2014

EXAMINERS REPORT ON THE PERFORMANCE OF CANDIDATES CSEE, 2014 THE NATIONAL EXAMINATIONS COUNCIL OF TANZANIA EXAMINERS REPORT ON THE PERFORMANCE OF CANDIDATES CSEE, 2014 062 BOOK KEEPING (For School Candidates) THE NATIONAL EXAMINATIONS COUNCIL OF TANZANIA EXAMINERS

More information

PRINCIPLES AND APPLICATIONS OF SPECIAL EDUCATION ASSESSMENT

PRINCIPLES AND APPLICATIONS OF SPECIAL EDUCATION ASSESSMENT PRINCIPLES AND APPLICATIONS OF SPECIAL EDUCATION ASSESSMENT CLASS 3: DESCRIPTIVE STATISTICS & RELIABILITY AND VALIDITY FEBRUARY 2, 2015 OBJECTIVES Define basic terminology used in assessment, such as validity,

More information

CHAPTER 8 T Tests. A number of t tests are available, including: The One-Sample T Test The Paired-Samples Test The Independent-Samples T Test

CHAPTER 8 T Tests. A number of t tests are available, including: The One-Sample T Test The Paired-Samples Test The Independent-Samples T Test CHAPTER 8 T Tests A number of t tests are available, including: The One-Sample T Test The Paired-Samples Test The Independent-Samples T Test 8.1. One-Sample T Test The One-Sample T Test procedure: Tests

More information

CHAPTER 10 REGRESSION AND CORRELATION

CHAPTER 10 REGRESSION AND CORRELATION CHAPTER 10 REGRESSION AND CORRELATION SIMPLE LINEAR REGRESSION: TWO VARIABLES (SECTIONS 10.1 10.3 OF UNDERSTANDABLE STATISTICS) Chapter 10 of Understandable Statistics introduces linear regression. The

More information

GETTING READY FOR DATA COLLECTION

GETTING READY FOR DATA COLLECTION 3 Chapter 7 Data Collection and Descriptive Statistics CHAPTER OBJECTIVES - STUDENTS SHOULD BE ABLE TO: Explain the steps in the data collection process. Construct a data collection form and code data

More information

Business Decisions. Spreadsheet Modeling. John F. Kros. Vfor\ B Third Edition Revised for Excel 2010

Business Decisions. Spreadsheet Modeling. John F. Kros. Vfor\ B Third Edition Revised for Excel 2010 1 Spreadsheet Modeling Vfor\ Business Decisions Third Edition Revised for Excel 2010 Featuring Risk Solver Platform for Decision Analysis, Simulation, and Optimization B12-1399 John F. Kros OffllEHS Preface

More information

THE LEAD PROFILE AND OTHER NON-PARAMETRIC TOOLS TO EVALUATE SURVEY SERIES AS LEADING INDICATORS

THE LEAD PROFILE AND OTHER NON-PARAMETRIC TOOLS TO EVALUATE SURVEY SERIES AS LEADING INDICATORS THE LEAD PROFILE AND OTHER NON-PARAMETRIC TOOLS TO EVALUATE SURVEY SERIES AS LEADING INDICATORS Anirvan Banerji New York 24th CIRET Conference Wellington, New Zealand March 17-20, 1999 Geoffrey H. Moore,

More information

Production Possibilities, Opportunity Cost, and Economic Growth

Production Possibilities, Opportunity Cost, and Economic Growth Chapter 2 Production Possibilities, Opportunity Cost, and Economic Growth CHAPTER SUMMARY The What, How and For Whom are introduced as the fundamental economic questions that must be addressed by all societies.

More information

FOUNDATIONCOURSE Syllabus. Fundamentals of Economics [50marks]

FOUNDATIONCOURSE Syllabus. Fundamentals of Economics [50marks] FOUNDATIONCOURSE Syllabus Section A: Fundamentals of Economics [50marks] 1. Basic Concepts of Economics a) The Fundamentals of Economics & Economic Organizations b) Utility, Wealth, Production and Capital

More information

Online Student Guide Types of Control Charts

Online Student Guide Types of Control Charts Online Student Guide Types of Control Charts OpusWorks 2016, All Rights Reserved 1 Table of Contents LEARNING OBJECTIVES... 4 INTRODUCTION... 4 DETECTION VS. PREVENTION... 5 CONTROL CHART UTILIZATION...

More information

Kvalitativ Introduktion til Matematik-Økonomi

Kvalitativ Introduktion til Matematik-Økonomi Kvalitativ Introduktion til Matematik-Økonomi matematik-økonomi studiet 1. basissemester Esben Høg I17 Aalborg Universitet 7. og 9. december 2009 Institut for Matematiske Fag Aalborg Universitet Esben

More information

INVESTIGATION OF SOME FACTORS AFFECTING MANUFACTURING WORKERS PERFORMANCE IN INDUSTRIES IN ANAMBRA STATE OF NIGERIA

INVESTIGATION OF SOME FACTORS AFFECTING MANUFACTURING WORKERS PERFORMANCE IN INDUSTRIES IN ANAMBRA STATE OF NIGERIA INVESTIGATION OF SOME FACTORS AFFECTING MANUFACTURING WORKERS PERFORMANCE IN INDUSTRIES IN ANAMBRA STATE OF NIGERIA Nwosu M. C., Ikwu G. O. R. and Uzorh A.C Department of Industrial and Production Engineering,

More information

CHAPTER 8 DESIGNING PAY LEVELS, MIX AND PAY STRUCTURES

CHAPTER 8 DESIGNING PAY LEVELS, MIX AND PAY STRUCTURES CHAPTER 8 DESIGNING PAY LEVELS, MIX AND PAY STRUCTURES LEARNING OBJECTIVES 1. Identify the major decisions in establishing externally competitive pay. 2. Describe the purpose(s) of a salary survey. 3.

More information

Managerial Economics Prof. Trupti Mishra S.J.M School of Management Indian Institute of Technology, Bombay. Lecture -29 Monopoly (Contd )

Managerial Economics Prof. Trupti Mishra S.J.M School of Management Indian Institute of Technology, Bombay. Lecture -29 Monopoly (Contd ) Managerial Economics Prof. Trupti Mishra S.J.M School of Management Indian Institute of Technology, Bombay Lecture -29 Monopoly (Contd ) In today s session, we will continue our discussion on monopoly.

More information

Control Charts for Customer Satisfaction Surveys

Control Charts for Customer Satisfaction Surveys Control Charts for Customer Satisfaction Surveys Robert Kushler Department of Mathematics and Statistics, Oakland University Gary Radka RDA Group ABSTRACT Periodic customer satisfaction surveys are used

More information

Contents in Brief. Preface

Contents in Brief. Preface Contents in Brief Preface Page v PART 1 INTRODUCTION 1 Chapter 1 Nature and Scope of Managerial Economics and Finance 3 Chapter 2 Equations, Graphs and Optimisation Techniques 21 Chapter 3 Demand, Supply

More information

Production Possibilities, Opportunity Cost, and Economic Growth

Production Possibilities, Opportunity Cost, and Economic Growth Chapter 2 Production Possibilities, Opportunity Cost, and Economic Growth CHAPTER SUMMARY The What, How and For Whom questions are introduced as the fundamental economic questions that must be addressed

More information

Priscilla Jennifer Rumbay. The Impact of THE IMPACT OF CUSTOMER LOYALTY PROGRAM TO CUSTOMER LOYALTY (STUDY OF GAUDI CLOTHING STORE MANADO)

Priscilla Jennifer Rumbay. The Impact of THE IMPACT OF CUSTOMER LOYALTY PROGRAM TO CUSTOMER LOYALTY (STUDY OF GAUDI CLOTHING STORE MANADO) THE IMPACT OF CUSTOMER LOYALTY PROGRAM TO CUSTOMER LOYALTY (STUDY OF GAUDI CLOTHING STORE MANADO) by: Priscilla Jennifer Rumbay 1 Faculty of Economics and Business International Business Administration

More information

PLANNING FOR PRODUCTION

PLANNING FOR PRODUCTION PLANNING FOR PRODUCTION Forecasting Forecasting is the first major activity in the planning, involving careful study of past data and present scenario to estimate the occurence, timing or magnitude of

More information

Choosing Smoothing Parameters For Exponential Smoothing: Minimizing Sums Of Squared Versus Sums Of Absolute Errors

Choosing Smoothing Parameters For Exponential Smoothing: Minimizing Sums Of Squared Versus Sums Of Absolute Errors Journal of Modern Applied Statistical Methods Volume 5 Issue 1 Article 11 5-1-2006 Choosing Smoothing Parameters For Exponential Smoothing: Minimizing Sums Of Squared Versus Sums Of Absolute Errors Terry

More information

JD Edwards World. Forecasting Guide Release A9.3 E

JD Edwards World. Forecasting Guide Release A9.3 E JD Edwards World Forecasting Guide Release A9.3 E20706-02 April 2013 JD Edwards World Forecasting Guide, Release A9.3 E20706-02 Copyright 2013, Oracle and/or its affiliates. All rights reserved. This software

More information

Chapter 2 Forecasting

Chapter 2 Forecasting Chapter 2 Forecasting There are two main reasons why an inventory control system needs to order items some time before customers demand them. First, there is nearly always a lead-time between the ordering

More information

The goods market. Screen 1

The goods market. Screen 1 The goods market Screen 1 In this presentation we take a closer look at the goods market and in particular how the demand for goods determines the level of production and income in the goods market. There

More information

Response Modeling Marketing Engineering Technical Note 1

Response Modeling Marketing Engineering Technical Note 1 Response Modeling Marketing Engineering Technical Note 1 Table of Contents Introduction Some Simple Market Response Models Linear model Power series model Fractional root model Semilog model Exponential

More information

College Name: Why most of the commercial fishing is dominated in higher latitude countries of the world?

College Name: Why most of the commercial fishing is dominated in higher latitude countries of the world? COMMERCIAL GEOGRAPHY; BA (BS) 311 BBA I Date: December 23, 2016 Max Time: 1.5 Hrs Max Marks: 30 1. Attempt any 5 questions. Do not write anything on the question paper 2. Mobile phone(s) or any other communicating

More information

Describing DSTs Analytics techniques

Describing DSTs Analytics techniques Describing DSTs Analytics techniques This document presents more detailed notes on the DST process and Analytics techniques 23/03/2015 1 SEAMS Copyright The contents of this document are subject to copyright

More information

Who Are My Best Customers?

Who Are My Best Customers? Technical report Who Are My Best Customers? Using SPSS to get greater value from your customer database Table of contents Introduction..............................................................2 Exploring

More information

Getting Started with OptQuest

Getting Started with OptQuest Getting Started with OptQuest What OptQuest does Futura Apartments model example Portfolio Allocation model example Defining decision variables in Crystal Ball Running OptQuest Specifying decision variable

More information

Chapter 12 Inventory Management. Inventory Management

Chapter 12 Inventory Management. Inventory Management Chapter 12 Inventory Management 2006 Prentice Hall, Inc. Outline Global Company Profile: Amazon.Com Functions Of Inventory Types of Inventory Inventory Management ABC Analysis Record Accuracy Cycle Counting

More information

Jacob: W hat if Framer Jacob has 10% percent of the U.S. wheat production? Is he still a competitive producer?

Jacob: W hat if Framer Jacob has 10% percent of the U.S. wheat production? Is he still a competitive producer? Microeconomics, Module 7: Competition in the Short Run (Chapter 7) Additional Illustrative Test Questions (The attached PDF file has better formatting.) Updated: June 9, 2005 Question 7.1: Pricing in a

More information

1.0 Chapter Introduction

1.0 Chapter Introduction 1.0 Chapter Introduction In this chapter, you will learn to use price index numbers to make the price adjustments necessary to analyze price and cost information collected over time. Price Index Numbers.

More information

Chapter 28 The Labor Market: Demand, Supply, and Outsourcing

Chapter 28 The Labor Market: Demand, Supply, and Outsourcing Chapter 28 The Labor Market: Demand, Supply, and Outsourcing Learning Objectives After you have studied this chapter, you should be able to 1. define marginal factor cost, marginal physical product of

More information

Accounting and Finance for Managers and Entrepreneurs

Accounting and Finance for Managers and Entrepreneurs Accounting and Finance for Managers and Entrepreneurs Course Description This course covers what everything business people and managers need to know about accounting and finance. It is directed toward

More information

COST ACCOUNTING QUESTION ONE

COST ACCOUNTING QUESTION ONE QUESTION ONE COST ACCOUNTING a) Briefly explain the term limiting factor A limiting factor is a binding constraint upon the organization. It prevents indefinite expansion or unlimited profits. It may be

More information

+? Mean +? No change -? Mean -? No Change. *? Mean *? Std *? Transformations & Data Cleaning. Transformations

+? Mean +? No change -? Mean -? No Change. *? Mean *? Std *? Transformations & Data Cleaning. Transformations Transformations Transformations & Data Cleaning Linear & non-linear transformations 2-kinds of Z-scores Identifying Outliers & Influential Cases Univariate Outlier Analyses -- trimming vs. Winsorizing

More information

Questionnaire. (3) (3) Bachelor s degree (3) Clerk (3) Third. (6) Other (specify) (6) Other (specify)

Questionnaire. (3) (3) Bachelor s degree (3) Clerk (3) Third. (6) Other (specify) (6) Other (specify) Questionnaire 1. Age (years) 2. Education 3. Job Level 4.Sex 5. Work Shift (1) Under 25 (1) High school (1) Manager (1) M (1) First (2) 25-35 (2) Some college (2) Supervisor (2) F (2) Second (3) 36-45

More information

Student Manual Principles of Economics

Student Manual Principles of Economics Student Manual Principles of Economics Beat The Market On Line: An Interactive Microeconomics Game. GoldSimulations www.goldsimulations.com Beat The Market : Student Manual Published by GoldSimulations

More information

Validation of a Multiple Linear Regression Model

Validation of a Multiple Linear Regression Model Validation of a Multiple Linear Regression Model Marin RUSĂNESCU 1, Anca Alexandra PURCĂREA 2 1 Valplast Industry Bucharest, rusanescum@yahoo.com 2 Politehnica University of Bucharest, Faculty of Entrepreneurship,

More information

Forecasting. Managerial Economics: Economic Tools for Today s Decision Makers, 4/e

Forecasting. Managerial Economics: Economic Tools for Today s Decision Makers, 4/e Forecasting Chapter 6 Managerial Economics: Economic Tools for Today s Decision Makers, 4/e By Paul Keat and Philip Young Forecasting Introduction Subjects of fforecasts Prerequisites for a Good Forecast

More information

Chapter 8 Designing Pay Levels, Mix, and Pay Structures

Chapter 8 Designing Pay Levels, Mix, and Pay Structures Chapter 8 Designing Pay Levels, Mix, and Pay Structures Major Decisions -Some major decisions in pay level determination: -determine pay level policy (specify employers external pay policy) -define purpose

More information

The Effects of Energy Efficient Heating Devices on Business Profit Maximization

The Effects of Energy Efficient Heating Devices on Business Profit Maximization , June 29 July 1, 2016, London, U.K. The Effects of Energy Efficient Heating Devices on Business Profit Maximization Patrick Onotu, Ijike D. Ona, and Uche M. DIMKPAGU, Member, IAENG Abstract Energy efficiency

More information

AMERICAN AGRICULTURE has become one of

AMERICAN AGRICULTURE has become one of AGRICULTURAL ECONOMICS RESEARCH Vol. 21, No. 1, JANUARY 1969 Technological Change in Agriculture By Robert 0. Nevel l AMERICAN AGRICULTURE has become one of the most productive industries in the world.

More information

Aggregate Planning and S&OP

Aggregate Planning and S&OP Aggregate Planning and S&OP 13 OUTLINE Global Company Profile: Frito-Lay The Planning Process Sales and Operations Planning The Nature of Aggregate Planning Aggregate Planning Strategies 1 OUTLINE - CONTINUED

More information

Hierarchical Linear Modeling: A Primer 1 (Measures Within People) R. C. Gardner Department of Psychology

Hierarchical Linear Modeling: A Primer 1 (Measures Within People) R. C. Gardner Department of Psychology Hierarchical Linear Modeling: A Primer 1 (Measures Within People) R. C. Gardner Department of Psychology As noted previously, Hierarchical Linear Modeling (HLM) can be considered a particular instance

More information

ECON 102 Kagundu Final Exam (New Material) Practice Exam Solutions

ECON 102 Kagundu Final Exam (New Material) Practice Exam Solutions www.liontutors.com ECON 102 Kagundu Final Exam (New Material) Practice Exam Solutions 1. A A large number of firms will be able to operate in the industry because you only need to produce a small amount

More information

IBM SPSS Forecasting 19

IBM SPSS Forecasting 19 IBM SPSS Forecasting 19 Note: Before using this information and the product it supports, read the general information under Notices on p. 108. This document contains proprietary information of SPSS Inc,

More information

More-Advanced Statistical Sampling Concepts for Tests of Controls and Tests of Balances

More-Advanced Statistical Sampling Concepts for Tests of Controls and Tests of Balances APPENDIX 10B More-Advanced Statistical Sampling Concepts for Tests of Controls and Tests of Balances Appendix 10B contains more mathematical and statistical details related to the test of controls sampling

More information

1.0 Introduction 1.1 Identifying Situations For Use 1.2 Constructing Price Index Numbers

1.0 Introduction 1.1 Identifying Situations For Use 1.2 Constructing Price Index Numbers 1.0 Introduction In this chapter, you will learn to use price index numbers to make the price adjustments necessary to analyze price and cost information collected over time. Price Index Numbers. Price

More information

ECON 2100 (Summer 2016 Sections 10 & 11) Exam #3C

ECON 2100 (Summer 2016 Sections 10 & 11) Exam #3C ECON 21 (Summer 216 Sections 1 & 11) Exam #3C Multiple Choice Questions: (3 points each) 1. I am taking of the exam. C. Version C 2. is a market structure in which there is one single seller of a unique

More information

15.4 Income Statement under Marginal Costing and Absorption Costing

15.4 Income Statement under Marginal Costing and Absorption Costing UNIT 15 Structure MARGINAL COSTING 15.0 Objectives 15.1 Introduction 15.2 Segregation of Mixed Costs 15.3 Concept of Marginal Cost and Marginal Costing 15.4 Income Statement under Marginal Costing and

More information

Short-Run Costs and Output Decisions

Short-Run Costs and Output Decisions Semester-I Course: 01 (Introductory Microeconomics) Unit IV - The Firm and Perfect Market Structure Lesson: Short-Run Costs and Output Decisions Lesson Developer: Jasmin Jawaharlal Nehru University Institute

More information

Forecasting for Short-Lived Products

Forecasting for Short-Lived Products HP Strategic Planning and Modeling Group Forecasting for Short-Lived Products Jim Burruss Dorothea Kuettner Hewlett-Packard, Inc. July, 22 Revision 2 About the Authors Jim Burruss is a Process Technology

More information

Chapter 10 Pure Monopoly

Chapter 10 Pure Monopoly Chapter 10 Pure Monopoly Multiple Choice Questions 1. Pure monopoly means: A. any market in which the demand curve to the firm is downsloping. B. a standardized product being produced by many firms. C.

More information

Chapter 3: The Relationship between Economic Freedom and Economic Well-Being

Chapter 3: The Relationship between Economic Freedom and Economic Well-Being 20 Economic Freedom of North America Chapter 3: The Relationship between Economic Freedom and Economic Well-Being A number of studies have linked levels of economic freedom with higher levels of economic

More information

Time Series Analysis in the Social Sciences

Time Series Analysis in the Social Sciences one Time Series Analysis in the Social Sciences in the social sciences, data are usually collected across space, that is, across countries, cities, and so on. Sometimes, however, data are collected across

More information

MULTIPLE CHOICE. Choose the one alternative that best completes the statement or answers the question. FIGURE 1-2

MULTIPLE CHOICE. Choose the one alternative that best completes the statement or answers the question. FIGURE 1-2 Questions of this SAMPLE exam were randomly chosen and may NOT be representative of the difficulty or focus of the actual examination. The professor did NOT review these questions. MULTIPLE CHOICE. Choose

More information

Chapter 13 Perfect Competition

Chapter 13 Perfect Competition Chapter 13 Perfect Competition 13.1 A Firm's Profit-Maximizing Choices 1) Is the number of sellers in the market the only thing that is different in each of the four market types economists study? Answer:

More information